Origins of alphavirus-derived small RNAs in mosquitoes

Abstract

The continual transmission in nature of many arthropod-borne viruses depends on the establishment of a persistent, nonpathogenic infection in a mosquito vector. The importance of antiviral immunity directed by small RNAs in the mechanism by which alphaviruses establish a persistent, nonpathogenic infection in the mosquito vector has recently been demonstrated. The origin of the small RNAs central to this RNA silencing response has recently been the subject of debate. Here we briefly summarize what is known about the mechanism of small RNA-directed immunity in invertebrates, and discuss current models for the viral triggers of this response. Finally, we summarize evidence indicating that alphavirus double-stranded replicative intermediates trigger an exogenous-siRNA pathway in mosquitoes resulting in the biogenesis of virus-derived siRNAs.

Introduction

RNA silencing is a process directed by small RNAs. Two major classes of small RNAs, short interfering RNAs (siRNAs) and microRNAs (miRNAs) are generated from double-stranded RNAs (dsRNAs) produced from a variety of sources. MicroRNAs are initially transcribed in the nucleus as primary miRNAs (pri-mRNAs) and then cleaved to generate precursor miRNAs (pre-miRNAs). The pre-miRNAs are exported from the nucleus to the cytoplasm where they are further processed to mature miRNAs. The siRNA pathway has recently been subdivided to reflect differences in the biogenesis of small RNAs from endogenous (e.g., transposable elements, convergent transcription and hairpin RNAs, etc.,) or exogenous (e.g., transgenes, viral replicative intermediates, etc.,) sources of dsRNA (reviewed in ref. 1). Evidence indicates that RNA silencing functions as a conserved antiviral defense in invertebrate organisms,² and plays an important role in the transmission of mosquito-borne viral diseases.³ Although a natural role for this pathway in antiviral defense is less certain in mammalian cells,⁴ several groups have demonstrated the feasibility of experimentally manipulating a mammalian RNA silencing pathway to interfere with virus replication.⁵ Thus, understanding the mechanism and origins of small RNA biogenesis in mosquitoes may ultimately result in the development of novel control strategies for arthropod-borne viral diseases. This article will summarize current knowledge about the origins of virus-derived small interfering RNAs (viRNAs) and the mechanism by which they direct an antiviral response in mosquitoes.

A Conserved Antiviral Response in Invertebrates Initiated by Dicer

Exogenous double-stranded precursors are processed by the RNase III enzyme Dicer into siRNA duplexes ~21–23 nucleotides in length.⁶ In Drosophila, siRNA duplexes are directed into an siRNA-induced silencing complex (siRISC) by a heterodimeric complex containing Dicer-2 (DCR-2) and the dsRNA-binding protein, R2D2.⁷ Argonaute 2 (AGO-2), an essential component of the siRISC, cleaves the passenger strand which is then ejected from the siRISC.^8–10 The remaining strand of the original siRNA duplex provides sequence specificity by serving as the guide strand for the active siRISC. Following Watson-Crick base pairing to a complementary RNA the “slicer” activity of AGO-2 cleaves the target, which results in further degradation of the fragmented RNA by other cellular nucleases.¹¹

Host Dicer proteins are at the vanguard of antiviral immunity directed by small RNAs due to a “sensor” role in detecting the dsRNA associated with viral replication. Drosophila melanogaster mutants possessing lesions in DCR-2 exhibit elevated levels of viral accumulation and an enhanced disease phenotype associated with increased mortality when infected with several different positive (+) strand RNA viruses.^12–14 The viruses used in these studies included several insect viruses: flock house virus (FHV), Drosophila C virus (DCV), cricket paralysis virus (CrPV); as well as a mosquito-borne alphavirus, Sindbis (SINV). The invertebrate Caenorhabditis elegans has a single Dicer which functions in the biogenesis of both miRNAs and siRNAs. Mutant worms possessing lesions in Dicer (DCR-1) exhibit developmental defects, presumably from disruption of the miRNA processing pathway.^15–17 Nevertheless, potentiation of virus replication in rde-1 and rde-4 null mutant worms using FHV and the arbovirus, vesicular stomatitis virus (VSV), have been demonstrated.^18,19 RDE-1 is a member of the argonaute family of proteins and RDE-4 is a dsRNA-binding protein.²⁰ RDE-1 and RDE-4 function together, and are thought to detect and present exogenous sources of dsRNA to DCR-1 for processing.²⁰ While the role of mammalian small RNA pathways in antiviral defense is less clear, the presence of a conserved antiviral silencing pathway in invertebrate organisms suggests that small RNAs may have an important role in the transmission of agents of human and animal disease.

Arboviruses are maintained in nature through biological transmission cycles that involve alternating virus replication in susceptible vertebrate and invertebrate hosts. Over 500 arboviruses have been cataloged, and more than 100 have been associated with human and animal diseases.²¹ In many arbovirus maintenance cycles, mosquitoes serve as the invertebrate vector species. Although infection of the vertebrate host is acute and often associated with disease, continual transmission of these viruses depends upon the establishment of a persistent, nonpathogenic infection in the mosquito vector. The more pathogenic the virus is in the mosquito, the more limited the probability for transmission. Thus, for the maintenance of arboviruses in nature it is imperative that very little fitness cost be associated with infections of the invertebrate host. An antiviral immune response directed by small RNAs has been shown to play a role in the mechanism by which the pathogenic potential of at least one important group of arboviruses is modulated in the mosquito.³ Infection of mosquitoes with alphaviruses triggers an antiviral RNA silencing response that results in the accumulation of viRNAs predominantly 21 nt in length (ref. ³ and Fig. 1A). When this RNA silencing response is suppressed in Aedes aegypti mosquitoes infected with SINV, using the B2 protein of FHV, viRNA production decreases resulting in elevated levels of virus replication and dramatically increased mortality, reminiscent of the enhanced disease phenotype associated with DCR-2 null mutant flies.³ Decreased survival of Anopheles gambiae after infection with recombinant ónyong-nyong virus (ONNV) expressing a suppressor of RNA silencing was also demonstrated.³ These results suggest that the accumulation of viRNAs restricts alphavirus replication and modulates pathogenesis in the mosquito host. Because suppressing viRNA-mediated restriction is associated with severe mortality in the disease vector, it is unlikely that alphaviruses encode suppressors of RNA silencing, or at least strong suppressors, as has been shown for a number of plant viruses and some insect viruses.²² These results also predict a prominent role for Dicer in sensing alphavirus replicative intermediates (RIs) in the cell. The B2 protein is required for FHV replication in cells possessing a fully functional siRNA pathway.^23,24 The structure of B2 suggests that it binds dsRNA regardless of length, and experimental results obtained in vitro show the protein inhibits DCR-2 processing of long dsRNA.^18,25–28 Thus, the previously described suppression of alphavirus-derived small RNAs by B2 likely results because DCR-2 is inhibited from processing dsRNA viral RIs present in the cell. Interestingly, depletion of DCR-2 in Ae. aegypti with dsRNA failed to increase mortality following infection with SINV.²⁹ Similarly, treatment of Drosophila S2 cells with dsRNA targeting DCR-2 failed to rescue the self-replication of a B2-deficient FHV RNA1, while replication was effectively rescued in dcr-2 null mutant embryos.^12,23 This may indicate that even reduced levels of DCR-2 are sufficient to initiate an RNA-silencing response capable of preventing mortality following alphavirus infection. The lack of mortality in the DCR-2-depleted mosquitoes might also be explained by functional redundancy in the mechanism of viRNA biogenesis. However, dcr-2 null mutant flies failed to accumulate detectable levels of viRNAs when infected with FHV, despite high level virus replication.¹² In contrast, abundant accumulation of viRNAs was detected in r2d2 null mutant flies infected with FHV.¹² Thus in Drosophila, DCR-2 appears to be essential to the generation of viRNAs. The presence of two Dicers in both the An. gambiae and Ae. aegypti genomes suggests that mosquitoes also subdivide processing of dsRNAs associated with the miRNA pathway (DCR-1) and dsRNAs associated with the siRNA pathway (DCR-2), but the question is more difficult to address in mosquitoes as null mutants in essential RNAi genes are not available. In fact, relatively few mosquito genes have been experimentally demonstrated to function in siRNA pathways. However, the recent development and validation of a transgenic mosquito strain that “senses” the status of the siRNA pathway will allow additional interrogation of mosquito genes and their role in RNAi.³⁰

Figure 1.

Consistent patterns of viRNA biogenesis following ONNV infection of An. gambiae and high-throughput sequencing from two independent biological replicates. Small RNA libraries from female An. gambiae mosquitoes (4 days post injection with ONNV) were constructed, sequenced and analyzed on an Illumina Genome Analyzer as described in Myles et al. (2008). After removing non-coding RNAs, we obtained 4,242,423 and 4,258,690 usable reads from biological replicates 1 and 2, respectively. Small RNAs matching the ONNV genome (100% match) were plotted by length (A) or were mapped onto the viral genome (B). Numbers of total 21 nt viRNAs (Total) or unique-mapping 21 nt viRNAs (Unique) deriving from the 5′ 2/3 (non-structural region) or 3′ 1/3 (structural region) of the viral genome are indicated for each biological replicate.

Origins of Virus-Derived Small RNAs in Inverterbrates

The overwhelming majority of arboviruses are RNA viruses classified into one of three major genera: the alphaviruses (Family Togaviridae), the flaviviruses (Family Flaviviridae) and the bunya-viruses (Family Orthobunyaviridae), all three of which have long been known to produce dsRNA during the course of their replication cycle in infected cells.^31–33 These dsRNA forms were first identified on the basis of their resistance to RNase A and were confirmed through the binding of dsRNA-specific antibodies.^34,35 SINV-derived dsRNA is produced in both vertebrate³⁵ and invertebrate cells.³⁴ Following initial identification of SINV dsRNA replicative forms (RFs) isolated from cells, their roles in virus replication were precisely defined by Simmons and Strauss.³⁶ The initial identification of dsRNA forms from dengue-2 virus,^32,35 Saint Louis encephalitis virus³⁷ and West Nile virus,³⁸ led to a model for the replication of flaviviruses through dsRNA intermediates as articulated by Cleaves et al.³⁹ Like alphaviruses, flavivirus negative (−) strands appear to be entirely incorporated in dsRNA structures throughout replication. LaCrosse virus (genus bunyavirus) has also been shown to form dsRNA structures during replication, as the 5′ and 3′ ends of each genomic segment are complementary, resulting in the circularization of the single-stranded segments.^33,40 The double-stranded region is estimated to be ~26–27 nt, and is present during infection of both vertebrate and invertebrate cells.⁴⁰ Thus, all arboviruses from these families would be expected to encounter an RNAi-based immune response upon infection of the invertebrate vector.

Imbalance in the synthesis of (−) and (+) strands is common in the infectious cycles of RNA viruses, usually with the genomic strand produced in greater abundance than its full-length complement. This forms the basis of a hypothesis in which viRNAs originate predominately from highly structured single-stranded viral RNAs.^41,42 Support for this hypothesis comes primarily from observations in which an excess of viRNAs were found to be derived from genomic RNAs in plants infected with several different (+) strand RNA viruses.^41,42 However, other (+) strand RNA plant viruses have been shown to generate viRNAs from both (+) and (−) strands in proportions more consistent with a predominantly dsRNA RI origin.^42,43 With regard to alphavi-ruses, small RNAs sequenced from infected mosquitoes reveal a significant bias for siRNAs derived from viral (+) strands (ref. ³; and Fig. 1A and B), but not in ratios that approximate imbalances in strand synthesis previously described for alphaviruses.⁴⁴

The genomic RNA of alphaviruses functions as the mRNA for translation of the viral nonstructural proteins and a template for the synthesis of a complementary (−) strand RNA. The viral (−) strand then serves as template for both the synthesis of new genomic-length (+) strand RNA and a shorter subgenomic length (+) strand RNA (26S mRNA) that encodes the viruses structural genes. Ribonuclease A digestion of SINV RIs produces three RFs possessing fully dsRNA cores.³⁶ RF I corresponds to the full-length virus genome in double-stranded form, while RFs II and III represent the nonstructural and structural regions, respectively (Fig. 1B-black bars). Thus, together RFs II and III also constitute the full-length genome in double-stranded form. The single-stranded (+)-sense genomic (corresponding to RF I) and subgenomic (corresponding to RF III) RNAs have been shown to be the predominant SINV-specific RNA species present in infected cells, with a single-stranded non-coding (+)-sense equivalent of RF II, RNA II, present in relatively lower quantities.^36,45 Cells infected with SINV have on average one single-stranded RNA II molecule to every 7 genomic and 50 subgenomic mRNA molecules.⁴⁵ Profiling of viRNAs isolated and sequenced from mosquitoes infected with either SINV or ONNV reveals no bias towards the region corresponding to the 26S mRNA (structural region) in each of the respective genomes, in contrast to what would be expected if viRNAs were originating from highly structured regions present in the more abundant single-stranded subgenomic mRNAs (ref. ³ and Fig 1B). Rather, as RF II and RF III (together the equivalent of RF I) have been shown to be present in equimolar amounts during the SINV infectious cycle these data suggest that the biogenesis of viRNA populations in mosquitoes occurs from the dsRNA SINV RIs.

Unlike viRNAs generated from FHV in S2 cells,^46,47 viRNAs derived from alphaviruses in mosquitoes do not appear to cluster from any particular region of the genome (ref. ³ and reviewed in Fig. 1B). While distributed across the entire genome, it is clear that some loci in the viral genome are hot spots for viRNA biogenesis (ref. ³ and reviewed in Fig. 1B). Consistent with biological replicate profiles of FHV in S2 cells,⁴⁷ the profiles of viRNAs in ONNV-infected mosquitoes are highly reproducible (Fig. 1B). Interestingly, the profile of viRNAs from ONNV-infected An. gambiae (Fig. 1B) is distinct from that of SINV-infected Ae. aegypti.³

Advances in our understanding of the RNAi pathway combined with an unprecedented availability of high-throughput sequencing technologies is enabling new investigations into the antiviral immune response of invertebrates. While to date high-throughput sequencing of viRNAs from animal virus infections has only been performed for FHV, SINV and ONNV,^3,46,47 many common questions have arisen. What is the biological explanation for viRNA hot spots, and what is their significance? Do abundant viRNAs represent the most important viRNAs in terms of modulating virus infection, are they decoys which once generated occupy the RNAi machinery but play no further role in modulation, or are these regions simply recognized preferentially by DCR-2? What is the underlying cause of the bias in (+) strand viRNAs over (−) strand viRNAs in ONNV and SINV infections of mosquitoes? Although results indicate dsRNA RIs are the primary substrates for DCR-2 in viRNA biogenesis, it remains possible that structured regions of the single stranded (+)-sense full-length genomic RNA, but not the more abundant subgenomic mRNA are targets of DCR-2. Future work characterizing the small RNA profiles from additional animal virus infections should shed light on these questions.

Abbreviations

viRNA

virus-derived small interfering RNA

RF

replicative form

RI

replicative intermediate